feat: add homepage and improve markdown processing

- Added an index route and template for the homepage - Moved to a custom markdown processor to handle relative URLs - Enhanced CSS for better layout structure and styling - Improved the directory listing logic to support path filtering - Updated endpoint to serve raw files from main branch - Added comprehensive documentation strings to Git class methods These changes improve user navigation and content display across the application.
2024-06-19 08:31:07 +02:00 · 2024-06-19 08:31:07 +02:00 · 621022dfb8
parent 6edc966f45
commit 621022dfb8
7 changed files with 160 additions and 31 deletions
--- a/src/gitcloak/app.py
+++ b/src/gitcloak/app.py
@ -1,6 +1,6 @@
 from flask import Flask, render_template, abort, send_from_directory
 from .classes.git import Git
-import markdown2
+from .classes.markdown import RelativeURLRewriter
 import logging
 from pathlib import Path

@ -14,6 +14,11 @@ logger.addHandler(handler)
 app = Flask(__name__)


+@app.route("/")
+def index():
+    return render_template("index.html")
+
+
@app.route("/assets/<path:path>")
 def send_assets(path):
    return send_from_directory(Path(__file__).parent / "assets", path)
@ -29,19 +34,23 @@ def get_tree(owner, repo, path=""):
    repo_url = f"https://github.com/{owner}/{repo}.git"
    git = Git(repo_url)
    try:
-        directory_structure = git.get_directory_structure()
-        filtered_structure = [
-            entry[len(path) :].lstrip("/")
-            for entry in directory_structure
-            if entry.startswith(path)
-        ]
+        directory_structure = git.get_directory_structure(path)
+        filtered_structure = directory_structure

        logger.debug(f"Filtered structure: {filtered_structure}")

        # Separate files and directories
-        directories = sorted(list(set(
-            [entry.split("/")[0] for entry in filtered_structure if "/" in entry]
-        )))
+        directories = sorted(
+            list(
+                set(
+                    [
+                        entry.split("/")[0]
+                        for entry in filtered_structure
+                        if "/" in entry
+                    ]
+                )
+            )
+        )
        files = [entry for entry in filtered_structure if "/" not in entry]

        # Get README.md content if it exists
@ -49,7 +58,10 @@ def get_tree(owner, repo, path=""):

        if f"README.md" in files:
            readme_md = git.get_file_content(f"{path}/README.md")
-            readme_content = markdown2.markdown(readme_md.decode("utf-8"))
+            base_url = f"/{owner}/{repo}/raw/main/{path}".rstrip("/")
+            readme_content = RelativeURLRewriter(base_url).convert(
+                readme_md.decode("utf-8")
+            )

        return render_template(
            "path.html",
@ -67,7 +79,7 @@ def get_tree(owner, repo, path=""):
        abort(404, description=str(e))


-@app.route("/<owner>/<repo>/raw/<path:file_path>", methods=["GET"])
+@app.route("/<owner>/<repo>/raw/main/<path:file_path>", methods=["GET"])
 def get_raw(owner, repo, file_path):
    repo_url = f"https://github.com/{owner}/{repo}.git"
    git = Git(repo_url)
--- a/src/gitcloak/assets/css/style.css
+++ b/src/gitcloak/assets/css/style.css
@ -5,6 +5,15 @@ body {
    height: 100%;
 }

+body {
+    display: flex;
+    flex-direction: column;
+}
+
+.container {
+    flex: 1;
+}
+
 .directory a,
 .file a {
    text-decoration: none;
@ -33,6 +42,9 @@ body {

 .icon {
    margin-right: 5px;
+    width: 1em;
+    height: 1em;
+    vertical-align: middle;
 }

 .list-group-item {
@ -43,4 +55,20 @@ body {
 .navbar-brand {
    font-size: 1.5rem;
    padding: 0 10px;
+}
+
+.footer {
+    position: fixed;
+    bottom: 0;
+    width: 100%;
+    text-align: center;
+}
+
+.text-white {
+    color: white;
+}
+
+.text-white a {
+    color: white;
+    text-decoration: underline;
 }
--- a/src/gitcloak/classes/git.py
+++ b/src/gitcloak/classes/git.py
@ -1,26 +1,67 @@
 import requests
-from dulwich.objects import Tree, Blob
+
+import logging
+
+from dulwich.objects import Tree, Blob, ShaFile, Tree
 from dulwich.client import HttpGitClient, get_transport_and_path
 from dulwich.repo import MemoryRepo

 class InMemoryRepo(MemoryRepo):
-    def get_tree(self, commit_sha):
+    def get_tree(self, commit_sha: bytes) -> Tree:
+        """Return the tree object for the given commit.
+
+        Args:
+            commit_sha (bytes): The commit hash.
+
+        Returns:
+            ShaFile: The tree object.
+        """
        commit = self.get_object(commit_sha)
        return self.get_object(commit.tree)

-    def list_tree(self, tree, prefix=""):
+    def list_tree(self, tree, path="", prefix=""):
+        """List the directory structure of the tree object.
+
+        Args:
+            tree (Tree): The tree object.
+            path (str): The path within the tree object.
+            prefix (str): The prefix to be added to the path.
+
+        Yields:
+            str: The path of the file or directory.
+        """
+        logging.debug(f"Listing tree {tree.sha()} with path {path}, prefix {prefix}")
+
        for entry in tree.items():
-            entry_path = f"{prefix}/{entry.path.decode('utf-8')}" if prefix else entry.path.decode('utf-8')
+            entry_path = (
+                f"{prefix}/{entry.path.decode('utf-8')}"
+                if prefix
+                else entry.path.decode("utf-8")
+            )
+
+            if path:
+                path_parts = path.split("/")
+                if path_parts[0] != entry.path.decode("utf-8"):
+                    continue
+
            if isinstance(self.get_object(entry.sha), Tree):
-                for _ in self.list_tree(self.get_object(entry.sha), entry_path):
-                    yield(_)
+                if path:
+                    for _ in self.list_tree(
+                        self.get_object(entry.sha), path="/".join(path_parts[1:]), prefix="/".join(path_parts[1:])
+                    ):
+                        yield (_)
+                else:
+                    for _ in self.list_tree(
+                        self.get_object(entry.sha), prefix=entry_path
+                    ):
+                        yield (_)
            else:
-                yield(entry_path)
+                yield (entry_path)

    def get_file_content(self, tree, file_path):
-        parts = file_path.split('/')
+        parts = file_path.split("/")
        for entry in tree.items():
-            entry_name = entry.path.decode('utf-8')
+            entry_name = entry.path.decode("utf-8")
            if entry_name == parts[0]:
                if len(parts) == 1:
                    file_obj = self.get_object(entry.sha)
@ -30,14 +71,17 @@ class InMemoryRepo(MemoryRepo):
                        raise ValueError(f"Path {file_path} is not a file.")
                else:
                    if isinstance(self.get_object(entry.sha), Tree):
-                        return self.get_file_content(self.get_object(entry.sha), '/'.join(parts[1:]))
+                        return self.get_file_content(
+                            self.get_object(entry.sha), "/".join(parts[1:])
+                        )
                    else:
                        raise ValueError(f"Path {file_path} is not a directory.")
        raise ValueError(f"File {file_path} not found in the repository.")

+
 class Git:
    def __init__(self, repo_url):
-        self.repo_url = repo_url.rstrip('/')
+        self.repo_url = repo_url.rstrip("/")
        self.client = HttpGitClient(self.repo_url)

    def get_remote_refs(self):
@ -46,16 +90,16 @@ class Git:
        return refs

    def get_head_commit(self, refs):
-        return refs[b'HEAD']
+        return refs[b"HEAD"]

    def get_pack_data(self, commit_sha):
        url = f"{self.repo_url}/git-upload-pack"
        request_body = f"0032want {commit_sha} multi_ack_detailed side-band-64k thin-pack ofs-delta agent=git/2.28.0\n00000009done\n"
-        response = requests.post(url, data=request_body.encode('utf-8'))
+        response = requests.post(url, data=request_body.encode("utf-8"))
        response.raise_for_status()
        return response.content

-    def get_directory_structure(self):
+    def get_directory_structure(self, path=""):
        # Initialize an in-memory repository
        self.repo = InMemoryRepo()

@ -67,10 +111,10 @@ class Git:
        tree = self.repo.get_tree(head_commit_hash)

        # List the directory structure
-        return list(self.repo.list_tree(tree))
+        return list(self.repo.list_tree(tree, path=path))

    def get_file_content(self, file_path):
-        file_path = file_path.lstrip('/')
+        file_path = file_path.lstrip("/")

        # Initialize an in-memory repository
        self.repo = InMemoryRepo()
@ -83,4 +127,4 @@ class Git:
        tree = self.repo.get_tree(head_commit_hash)

        # Get the file content
-        return self.repo.get_file_content(tree, file_path)
+        return self.repo.get_file_content(tree, file_path)
--- a/src/gitcloak/classes/markdown.py
+++ b/src/gitcloak/classes/markdown.py
@ -0,0 +1,20 @@
+import re
+import markdown2
+
+class RelativeURLRewriter(markdown2.Markdown):
+    def __init__(self, base_url, *args, **kwargs):
+        self.base_url = base_url
+        super().__init__(*args, **kwargs)
+
+    def postprocess(self, text):
+        # Rewrite relative URLs
+        def replace_url(match):
+            url = match.group(1)
+            print(f"URL: {match}")
+            if not (":" in url or url.startswith("/") or url.startswith("#") or url.startswith("md5-")):
+                return f'src="{self.base_url}/{url}"'
+            return match.group(0)
+
+        text = re.sub(r'src="([^"]+)"', replace_url, text)
+        text = re.sub(r'href="([^"]+)"', replace_url, text)
+        return text
--- a/src/gitcloak/templates/base.html
+++ b/src/gitcloak/templates/base.html
@ -8,11 +8,16 @@
    <link href="/assets/css/style.css" rel="stylesheet">
 </head>
 <body>
-    <nav class="navbar navbar-expand-lg navbar-dark bg-dark">
+    <nav class="navbar navbar-dark bg-dark">
        <a class="navbar-brand" href="/">GitCloak</a>
    </nav>
    <div class="container mt-4">
        {% block content %}{% endblock %}
    </div>
+    <footer class="footer mt-auto py-3 bg-dark text-white">
+        <div class="container">
+            <span>GitCloak is brought to you by <a href="https://private.coffee">Private.coffee</a>.</span>
+        </div>
+    </footer>
 </body>
 </html>
--- a/src/gitcloak/templates/index.html
+++ b/src/gitcloak/templates/index.html
@ -0,0 +1,20 @@
+{% extends 'base.html' %}
+
+{% block title %}Welcome to GitCloak{% endblock %}
+
+{% block content %}
+<div class="jumbotron">
+    <h1 class="display-8">Welcome to GitCloak!</h1>
+    <p class="lead">GitCloak is a simple web frontend that lets you browse public GitHub repositories and view their contents.</p>
+    <p class="lead">Unlike other GitHub interfaces, GitCloak does not use APIs or screen scraping, needs no credentials and does not use any JavaScript.</p>
+    <hr class="my-4">
+    <p>To get started, simply replace <code>github.com</code> in the URL - the following paths are already available:</p>
+    <ul>
+        <li><code>/&lt;owner&gt;/&lt;repo&gt;/</code> - View the root directory of the repository</li>
+        <li><code>/&lt;owner&gt;/&lt;repo&gt;/tree/main/&lt;path&gt;</code> - View a specific directory</li>
+        <li><code>/&lt;owner&gt;/&lt;repo&gt;/raw/&lt;file_path&gt;</code> - View the raw content of a specific file</li>
+    </ul>
+    <p>For example, to see the contents of the <code>PrivateCoffee/transfer.coffee</code> repository, simply visit <a href="/PrivateCoffee/transfer.coffee/">/PrivateCoffee/transfer.coffee/</a>.</p>
+    <p>Other features are still in development, so stay tuned!</p>
+</div>
+{% endblock %}
--- a/test.py
+++ b/test.py
@ -2,7 +2,7 @@ from gitcloak.classes.git import Git

 repo = "https://github.com/privatecoffee/transfer.coffee"
 git = Git(repo)
-print(git.get_directory_structure())
+print(git.get_directory_structure("public/"))

 file_path = "public/dist/js/webtorrent.LICENSE"
 content = git.get_file_content(file_path)